Executive Summary : | This proposal aims to address the driver dilemma detection problem by using physiological measurement techniques like EEG, which offer low latency and accurate solutions. However, increasing the number of EEG channels can lead to increased accuracy in drowsiness detection but also increases processing costs. The 32-channel EEG systems are uncomfortable for subjects and not suitable for practical applications. There is a need for a fatigue detection system that operates with minimal channels and is comfortable for users. Advanced signal processing and machine learning tools with faster convergence rates are needed for optimal time processing in prediction systems. OpenBCI, a low-cost EEG device, is used for BCI training, with EEGNet for advanced signal processing and HJB-based optimal machine learning method for optimal convergence. The prototype will feature deep learning-based deciphering of EEG signals in terms of fatigue level, attention, and stress levels. A real-time cognitive state detection module will be assembled, with predicted cognitive states visible through a webapp or android app. A recommendation system for deferring driving or auto-piloting can be developed. The goal is to minimize the number of EEG channels and develop optimal machine learning methods and novel data science paradigms to create a workable real-time cognitive state monitoring Brain Computer Interface that can be used by individuals from various fields, including students, security officers, plant operators, drivers, and air pilots. |